Multiple mutations of the human cytochrome P450IID6 gene (CYP2D6) in poor metabolizers of debrisoquine. Study of the functional significance of individual mutations by expression of chimeric genes

Abstract

The debrisoquine/sparteine-type polymorphism is a clinically important inherited variation of drug metabolism characterized by two phenotypes, the extensive metabolizer and the poor metabolizer (PM). Five to 10 percent of individuals in Caucasian populations are of the PM phenotype and have deficient metabolism of debrisoquine and over 25 other drugs. Our previous studies have revealed absence of cytochrome P450IID6 protein and aberrant splicing of IID6 premRNA in livers of PMs. Moreover, two mutant alleles of the P450IID6 gene locus (CYP2D6) were identified by restriction fragment length analysis to be associated with the PM phenotype. However, the mutations of the CYP2D6 gene causing absent P450IID6 protein have not been defined. Here we report the cloning and sequencing of two types of mutant alleles of CYP2D6 isolated from genomic libraries of three PM individuals. One allele (29-A) was characterized by a single nucleotide deletion in the 5th exon with consequent frameshift and was observed in one individual only. The other type of mutant allele (29-B) was present in all three PM individuals and its sequence contained multiple mutations, notably four base changes causing amino acid changes in exons 1, 2 and 9, and a point mutation at the consensus sequence of the splice site of the 3rd intron. To understand the significance of the individual mutations, chimeric genes were constructed between the wild-type IID6 gene and the mutant 29-B allele or site-specific mutations were introduced into the IID6-cDNA and these DNA constructs were transiently expressed in COS-1 cells. The mutations in exon 1 resulted in a functionally deficient IID6 protein and the mutation at the splice site in absent IID6 protein, whereas the mutations in exons 2 and 9 were of no consequence for IID6 function. Only the mutation at the splice site thus explains the absence of P450IID6 protein in livers of PM individuals and appears to be a common cause of polymorphic drug oxidation.

Full text

THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 265, No. 26, Issue of October 5. pp. l-7209-17214,1990
c 1990 by The American Society for Biochemistry and Molecular Biology, Inc.
Printed in CJ S. A.
Multiple Mutations of the Human Cytochrome P450IID6 Gene
(CYP2D6) in Poor Metabolizers of Debrisoquine
STUDY OF THE FUNCTIONAL SIGNIFICANCE OF INDIVIDUAL MUTATIONS BY EXPRESSION
OF CHIMERIC GENES*
(Received for publication, April 20, 1990)
Masaaki Kagimoto$., Markus Heim, Keiko Kagimotog, Tanja Zeugin, and Urs A. Meyer
From the Department of Pharmacology, Biocenter of the Uniuersity of Base& CH-4056 Base& Switzerland
The debrisoquine/sparteine-type polymorphism is a
clinically important inherited variation of drug metab-
olism characterized by two phenotypes, the extensive
metabolizer and the poor metabolizer (PM). Five to 10
percent of individuals in Caucasian populations are of
the PM phenotype and have deficient metabolism of
debrisoquine and over 25 other drugs. Our previous
studies have revealed absence of cytochrome P450IID6
protein and aberrant splicing of IID premRNA in
livers of PMs. Moreover, two mutant alleles of the
P450IID6 gene locus (CYP2D6) were identified by
restriction fragment length analysis to be associated
with the PM phenotype. However, the mutations of the
CYP2D6 gene causing absent P450IID6 protein have
not been defined.
Here we report the cloning and sequencing of two
types of mutant alleles of CYP2D6 isolated from ge-
nomic libraries of three PM individuals. One allele (29-
A) was characterized by a single nucleotide deletion in
the 5th exon with consequent frameshift and was ob-
served in one individual only. The other type of mutant
allele (29-B) was present in all three PM individuals
and its sequence contained multiple mutations, notably
four base changes causing amino acid changes in exons
1, 2 and 9, and a point mutation at the consensus
sequence of the splice site of the 3rd intron. To under-
stand the significance of the individual mutations, chi-
merit genes were constructed between the wild-type
IID gene and the mutant 29-B allele or site-specific
mutations were introduced into the IIDG-cDNA and
these DNA constructs were transiently expressed in
COS-1 cells. The mutations in exon 1 resulted in a
functionally deficient IID protein and the mutation at
the splice site in absent IID protein, whereas the
mutations in exons 2 and 9 were of no consequence for
IID function. Only the mutation at the splice site thus
explains the absence of P450IID6 protein in livers of
PM individuals and appears to be a common cause of
polymorphic drug oxidation.
The debrisoquine/sparteine-type polymorphism of drug ox-
idation is one of the most extensively studied genetically
* This research was supported by Grant 3.817.87 from the Swiss
National Science Foundation. The costs of publication of this article
were defrayed in part by the payment of page charges. This article
must therefore be hereby marked “aduertisement” in accordance with
18 U.S.C. Section 1734 solely to indicate this fact.
$ Present address: First Dept. of Internal Medicine, Faculty of
Medicine, Kyushu University, Fukuoka 812, Japan.
§ Present address: Dept. of Plastic
Surgery, School of Medicine,
Kurume University, Kurume-shi, 830, Japan.
determined variations in drug metabolism (1, 2). It causes
deficient metabolism of debrisoquine, sparteine, bufuralol,
dextromethorphan, and numerous other drugs in so-called
poor metabolizer (PM)’ individuals. The PM phenotype is
inherited as an autosomal recessive trait and occurs with a
frequency of 5-10% in most populations studied (3).
Our earlier investigations have shown that defective metab-
olism of drugs is due to the absence of cytochrome P450IID6’
in the liver of PM individuals with the PM phenotype (4).
Moreover, cDNA analysis of RNA from two PM livers pro-
vided evidence for incorrectly spliced pre-mRNAs as a possi-
ble cause for absent IID protein (4, 5). The gene for
P450IID6, designated CYP2D6 (6), has been localized to
chromosome 22 (7). A presumed pseudogene CYP2D7 and a
definite pseudogene CYP2D8 are localized 5’ of the CYP2D6
locus (8). In further studies, two mutant alleles of the CYP2D6
gene were identified by restriction fragment length analysis
of genomic DNA after hybridization with the IID cDNA (9).
These mutant alleles are reflected by 44- and 11.5-kb frag-
ments after digestion with the XbaI endonuclease and, when
present together, are linked to the PM phenotype. However,
the presence of the two mutant alleles allows prediction of
the phenotype in only 25% of PMs. Additional gene-inacti-
vating mutations therefore must be present in 75% of PM
individuals in which no or only one mutant allele can be
identified by restriction fragment length polymorphism
(RFLP). Both the RFLPs and the cDNA analysis of RNA
from PM livers therefore suggest that multiple mutations can
cause the PM phenotype. However, neither the DNA muta-
tions causing the incorrect splicing of IID pre-mRNA nor
the mutations responsible for the RFLP, nor other mutations
of the CYP2D6 gene which can explain the absence of
P450IID6 protein, have been identified.
To characterize the mutations causing absent P450IID6
protein in PMs, clones containing the CYP2D6 genes were
isolated from genomic libraries constructed from leukocyte
DNA of three PM individuals, all belonging to the group in
which no mutant allele can be detected by RFLP analysis.
Two types of mutant alleles were isolated, their exons and
exon-intron junctions were sequenced and the entire gene was
inserted into the pCMV vector and transiently expressed in
cell culture. The functional significance of the different mu-
1 The abbreviations used are: PM, poor metabolizer; P450IID6,
cytochrome P450IID6; RFLP, restriction fragment length polymor-
phism; DMEM, Dulbecco’s modified Eagle’s medium; bp, base pairs;
kb, kilobase pairs.
‘The cytochrome P-450 enzyme which is deficient in PMs of
debrisoquine has previously been called P450bufl (4), P45Odbl (5),
and P450IIDl (7). P450IID6 is the designation proposed in the most
recent update on P-450 nomenclature (6) and is used throughout this
paper.
17209
This is an Open Access article under the CC BY license.

17210 Debrisoquine Polymorphism
tations in one of these alleles was tested by construction of
chimeras between the mutant and wild-type IID genes or by
introducing mutations into the wild-type cDNA. The chimeric
proteins which were produced by expression of these con-
structs in COS-1 cells were analyzed by immunological tech-
niques
and functional assays. Mutations which cause deficient
IID protein and mutations without functional consequences
for debrisoquine metabolism could thereby be separated.
MATERIALS AND METHODS
Characterization
of
PM Individuals
The leukocyte DNA of three individuals of PM phenotype was
selected for this study from our previous collection of DNA samples
for population and family studies (9). The three subjects indexed as
PM1 (ZICL), PM2 (KABI), and PM3 (B07) were identified as PMs
by phenotyping with either debrisoquine (10) or sparteine (11). The
urinary debrisoquine/4-OH-debrisoquine metabolic ratio was 332 and
55 for PM1 and PM2, respectively. PM3 was identified with the
urinary sparteine/dehydrosparteine metabolism ratio, which was 250.
These PMs were selected because they are of the XbaI 29/29-kb
genotype, which provides no information on the phenotype by restric-
tion analysis. The sequence information of the wild-type CYP2D6
gene used for comparison was from a homozygous extensive metabo-
lizer individual (EZA) recently described (8). Southern blot analysis
was performed as described previously (9).
Cloning and Sequencing
The strategy of cloning was based on the information derived from
Southern blots with BamHI, EcoRI, and XbaI and the almost com-
plete sequence of the area of the three genes CYP2D6, CYP2D7, and
CYP2D8 isolated from extensive metabolizer DNA (8, 9). According
to this, the 16-kb EcoRI fragment contains the CYP2D7 gene, which
is located 5’ of the normal CYP2D6. The 9.4-kb EcoRI fragment
represents the CYP2D6 gene and the 8.5kb fragment the CYP2D8
pseudogene. Genomic libraries were constructed from leukocyte DNA
of each of the three PM individuals. Since Southern blot analysis of
their genomic DNA had the same EcoRI pattern as EMS it was
assumed that their 16-, 9.4-, and 8.5-kb fragments contain the same
CYPZD genes as DNA of the wild-type or homozygous extensive
metabolizer. DNA was completely digested with EcoRI and inserted
into the vector XgtWES (Bethesda Research Laboratories; Ref. 12).
This vector can process 2%15-kb inserts. Because the 16-kb fragment
is too long to be accepted by XgtWES, these libraries contain only
the 9.4- and 8.5-kb fragments, corresponding to the CYP2D6 and
CYP2D8 genes. The libraries were screened with two probes to ensure
the identification of clones representing the CYP2D6 gene. Both
probes were labeled with ‘*P by nick translation. The first screening
was done with the full length IIDG-cDNA (5), which recognizes both
the CYP2D6 and CYP2D8 clones. We therefore used an additional
probe, a Sac1 0.4-kb fragment (bp -717 to -305) prepared from the
genomic DNA of a homozygous extensive metabolizer. This fragment
recognizes the 5’-flanking region of both the CYP2D6 and CYP2D7
gene, but not the CYP2D8 gene. As the library only contains CYP2D8
and CYP2D6 genes, the second screening with this probe thus selects
for CYP2D6. The EcoRI fragments of the positive clones were di-
gested with various restriction enzymes to smaller DNA fragments;
these were subcloned into pUCI9 and sequenced by the double strand
dideoxy chain termination method (13, 14), using universal and
reverse primers as well as 18 synthesized oligonucleotides (20-mers)
corresponding to the 5’ and 3’ part of each of the 9 exons and the
intron-exon junctions.
Construction
of
Expression Clones
The construction of the full length expression clones is summarized
in Fig. 1.
CYP2D6 Wild-type-An AccI-KpnI fragment from the 3’ part of
CYP2D6 wild-type gene was blunt-ended by treatment with T4 DNA
polmyerase (BRL) and subcloned into pUC19 by using the HincII
site in the correct orientation. The HincII fragment of the same gene
was subcloned into the SmaI site of the Bluescript vector (Promega)
in the correct orientation. The EcoRI-BamHI fragment of the former
clone was then replaced with that of the latter clone to construct a
full length gene in pUC19. The resulting gene was excised by EcoRI
UP206
Iwild type1
29 - B
FIG. 1. Schematic description of the construction of a eu-
karyotic expression vector containing the wild-type and chi-
merit CYP2D6 genes. The procedure and the vector are detailed
under “Materials and Methods.” E, EcoRI; B, BamHI; Hc, HincII;
Bs, BssHII; A, AccI; K, KpnI; S, SmaI; H, HindIII. The chimeric
genes of Fig. 4A were assembled in pUC19 using combinations of the
three parts EcoRI-BssHII (1.8 kb), BssHII-AccI (0.8 kb), and AccI-
Hind111 (1.8 kb).
and Hind111 and inserted into the pCMV expression vector (15, 16),
using the same restriction sites.
CYP2D6 Mutated Genes (29-B Allele)-The EcoRI-KpnI fragment
of the mutant 29-B gene was subcloned into pUC19. This clone was
digested by Hind111 and SalI, blunt-ended by T4 DNA polymerase,
and ligated again to eliminate the AccI site in the vector. The Hind111
site in the vector was maintained during this procedure. The HincII
fragment of the mutated 29-B gene was then subcloned into the SmaI
site of another pUC19 in the correct orientation and the EcoRI-
BssHII fragment of the former clone was replaced with that of this
clone. The engineered full length gene in pUC19 was further sub-
cloned into pCMV using EcoRI and Hind111 sites.
Chimeric Genes
Chimeric genes were assembled in pUC19 using combinations of
the three parts (EcoRI-BssHII, 1.8 kb; BssHII-AccI, 0.8 kb; AccI-
HindIII, 1.8 kb) of the constructed full length gene clones (Fig. 4A).
The total length of the chimeric genes thus was 4.4 kb. The chimeric
genes were inserted as described above into pCMV using EcoRI and
Hind111 restriction sites.
cDNAs with Point Mutations
A full length human CPY2D6-cDNA was constructed by subclon-
ing a 400-bp EcoRI-SmaI fragment containing the first 140 bp of the
coding sequence and 260 bp of the 5’-untranslated region of a
CYP2D6 wild-type genomic clone (8) into a rat-human hybrid cDNA
that was deleted of the corresponding part by cutting it with the same
restriction enzyme. The same strategy was used to construct a cDNA
containing only mutation 1 (MI, 188 C to T) by using a genomic
clone of a 29-B allele. Mutation 2 (MII, 1062 C to A) and mutation 3
(MIII, 1072 A to G) were introduced into the wild-type cDNA by the
polymerase chain reaction according to Kammann et al. (17), using
two mutagenic primers (GGGTCACCATCGCCTCGCG for MII,
TCCTCGCCGCGGGTCACCA for MIII). A unique XhoII site was
used to subclone the polymerase chain reaction generated fragments
into the wild-type cDNA (Fig. 5). All constructs were sequenced to
exclude polymerase chain reaction artifacts.

Debrisoquine Polymorphism 17211
DNA Transfection of COS-1 Cells
Expression clones were transfected into COS-1 cells (18) by the
DEAE-dextran method (19, 20) with slight modifications. Sixteen
hours before transfection, COS-1 cells were passaged from a confluent
loo-mm culture dish to four dishes in Dulbecco’s modified Eagle’s
medium (DMEM) containing 10% fetal calf serum. Transfection was
performed by incubation of the cells for 2 h with DEAE-dextran (250
pg/ml; Pharmacia) and DNA (20 fig/plate) in serum-free DMEM,
followed by an incubation for 3 h in DMEM containing 10% fetal
calf serum and chloroquine (52 pg/ml; Sigma). The cells were har-
vested for analysis of IID protein and function after 66 h of incu-
bation in DMEM with 10% fetal calf serum. For an assessment of
IID function in intact cultured cells, (+)-bufuralol (200
pM)
was
added to the cultures for the last 24 h, and l’-hydroxybufuralol
analyzed in the medium (21).
RNA Blot Analysis
Twenty micrograms of total RNA, which was isolated (22) from
the transfected COS-1 cells, was size-fractionated by electrophoresis
in 1.0% agarose-formaldehyde gels (23). The full length CYP2D6
cDNA, which was radiolabeled by the random priming method (24),
was used as the probe. Transfer of the RNA to a nylon membrane
(GeneScreen Plus; DuPont-New England Nuclear) and hybridization
with the radiolabeled probe were performed under the conditions
recommended by Du Pont.
Immunoblot Analysis
Sodium dodecyl sulfate-polyacrylamide gel electrophoresis of COS-
1 cell homogenates (protein 50-100 pg/lane) was performed in a 10%
polyacrylamide gel, and the proteins were transferred to nitrocellulose
and then exposed to the monoclonal antibody 114/2 and subsequently
to rabbit anti-mouse IgG. The bound IgG was visualized by auto-
radiography after incubation with “‘I-protein A. Details of the tech-
nique and the specificity of monoclonal antibody 114/2 in the recog-
nition of P450IID6 have recently been described (21).
Assay of Bufuralol 1 ‘-Hydroxylation
Bufuralol l’-hydroxylation assays were done as described for mi-
crosomal fractions (4) in the presence of NADPH and 02. COS-1
cells were harvested in phosphate-buffered saline, the suspension
centrifuged at 1000
x
g for 3 min, the pellet resuspended in sodium
phosphate buffer, pH 7.4, sonicated 3 times for 10 s at 4 “C, and the
assay performed with 350 pg of protein. Substrate concentration of
(+)-bufuralol was 500
FM.
RESULTS
Cloning
and Sequencing
of
Mutant CYP2D6 Genes
Two positive clones were isolated
from the genomic libraries
of each of the three PM individuals (Fig. 2). Four of these six
clones were fully sequenced in all exons and intron-exon
junctions. The remaining two clones were only partially se-
quenced as detailed below.
PM1 and PM2-One
clone of the two isolated from each
PM library was first sequenced in all exons and intron-exon
junctions. It became clear that these two clones had identical
mutations as well as an additional BamHI restriction site
when compared with the wild-type CYP2D6 gene (Figs. 2 and
3). They were designated 29-B. The second clone from each
PM was sequenced only in two areas where mutations had
been identified, namely the 3’-intron-exon junction of the 3rd
intron and exon 2. The same mutations were again detected
as well as the additional BamHI site present in both alleles
in PM1 and PM2, already evident in the genomic Southern
blot analysis. It is of course unknown whether or not the two
clones are derived from the same allele.
The mutations of the 29-B allele are summarized in Fig. 3.
They include two silent mutations (1085 C to G, 1749 G to
C), four amino acid changes (188 C to T resulting in 34 Pro
to Ser, 1062 C
to
A resulting in 91 Leu to Met, 1072 A to G
resulting in 94 His
to
Arg, 4268 G to C resulting in 486 Ser
to Thr), and one nucleotide change (1934 G to A) at the 3’
;L--iGL= 29-B
PM1
li (29-B)
i--ah-d 29-B
PM2
(29-B)
29-B’
PM3
u 29-A
Ikb
FIG. 2. Restriction analysis of genomic DNA clones of three
poor metabolizers of debrisoquine. Two clones from each indi-
vidual containing the 9.4-kb EcoRI fragment with the CYP2D6 gene
were analyzed.
E, EcoRI; B, BamHI. The star indicates the additional
BamHI site found in the alleles designated 29-B and 29-B’; all the
exons and intron-exon junctions were sequenced, except for the two
clones (29-B), in which only the areas with mutations were sequenced.
(A) 1
E B
29-A L 1
B E
17 n nn nn
1
2 36 567 89
(PW)C
E 6
29-B 1 1
B E
I
(Mel) :
CArgIG G (Thr) C
9 * f
(LedC (His)A t (Ser) t
FIG. 3. Localization of mutations of two alleles (29-A, 29-
B) of the CYP2D6 gene. Star indicates the additional BamHI site
(B). The exact locations of all mutations are given in the text.
end of the 3rd intron. The G to A change at the acceptor site
consensus sequence presumably results in incorrect splicing,
but the 29-B allele contains multiple additional mutations
and may have more in the unsequenced introns. The sequence
positions correspond to the recently published CYP2D6 se-
quence (8).
PM3-With the knowledge of the additional BamHI site
in the clones from PM1 and PM2, PM3 was selected to be
studied because the poor metabolizer individual in his genomic
Southern blot was heterozygous for this BamHI site. Both
clones from PM3 were fully sequenced. One allele was iden-
tical to the mutant allele 29-B, except for the silent mutation
(1749 G to C) in exon 3 and it therefore was designated 29-
B’. The other allele, which had no additional BamHI site,
had one nucleotide deletion (2637 A) in the 5th exon resulting
in a frame shift. This allele was designated 29-A (Figs. 2 and
3).
Expression
of
the CYP2D6 Wild-type Gene, the Mutated
29-B Allele, and Chimeras
of
Both Genes
To evaluate the functional consequences of each mutation
and of unsequenced introns, we constructed full length gene
expression clones of CYP2D6 wild-type and the mutated 29-
B allele (Fig. 2). No difference was found in the 5’-flanking

17212 Debrisoquine Polymorphism
sequences (bp -77 to start codon) of the 29-B and CYP2D6
genes and no additional translation start signal (ATG) was
present in this area. After construction of these clones, we
divided the two genes into three parts and exchanged these
parts in order to construct the four chimeric genes shown in
Fig. 4-4. The middle fragment of 29-B was completely se-
quenced and a mutation (2185 A to G) was detected in intron
4 in addition to the two previously identified mutations in
this segment, namely the one silent mutation in the 3rd exon
and the G to A mutation of the last nucleotide of the 3rd
intron (Fig. 3).
CYP2D Wild-type Gene and Mutated 29-B Allele-As
shown in Fig. 4B, the CYP2D6 wild-type gene construct
produced functional and immunoreactive protein in COS-1
cells. The mutated 29-B gene on expression did not result in
recognizable protein and no enzymatic activity could be dem-
onstrated in transfected COS-1 cells. Interestingly, mRNA
was recognized by the IID cDNA in Northern blots and had
the same apparent size for both constructs.
Chimeric Genes 1 and 2-Because of the suspected impor-
tance of the mutation in the splice-site consensus sequence
at the 3’ end of the 3rd intron, we first constructed chimeric
genes which would allow us to test the consequences of this
mutation. The chimeric gene 1 includes the middle part of
the mutated 29-B allele, and the 5’ and 3’ part of the wild-
type gene (Fig. 4A). In construct 2, on the other hand, the
middle part was derived from the wild-type gene, and the 5’
and 3’ part from the mutated 29-B allele. On expression,
chimeric gene 1 resulted in no recognizable protein and no
enzyme activity, as with the entire 29-B gene. The size of the
mRNA was identical to the 29-B and wild-type gene products.
Chimeric gene 2 apparently produced a similar amount of
IID protein as the wild-type IID gene, but no significant
activity over COS-1 cell controls.
Chimeric Genes 3 and ~--TO evaluate the effect of the
amino acid changes in the 5’ and 3’ part of the 29-B allele,
we constructed clones 3 and 4 which include the 5’ and 3’
parts of the 29-B gene, each combined with the other two
parts of the wild-type gene. Clone 3 produced an immuno-
reactive protein, but no activity, as did chimeric gene 2
A
5’
Bs A
1 - w
;B; W
I I
2
i i J
B
;W; B
3
,- i/ ,
B ;w: w
I ,
L i i
L J
w w B
I’-OH-bulurolol
5.0
Activity
FIG.
4. Expression of chimeric gene constructs from wild-
type CYP2D6 and its mutant 29-B allele in CO&l cells.
A,
description of the three parts of the wild-type (
W)
gene which were
exchanged with the corresponding parts of the 29-B allele (i?) with
mutations. 0, the mutations causing amino acid changes in exons 1,
2, and 9;
X,
the mutation in the splice site consensus sequence of the
3rd intron.
B,
Northern blot
(nRNA),
Western blot (Protein), and
bufuralol l’-hydroxylation of COS-1 cell extracts 66 h after transfec-
tion with the DNA constructs
l-4,
the intact wild-type (CYp2D6),
and the mutant
(29-B)
gene. Control, mock-transfected cells.
cDNA
EC0 RI Smal
XhoU Hnnd III
5’ ATG
-1Vsl3* 100 271
MI Mu %I
C-T C-A A-G
EXPRESSION
Protein
wt MI Mu Mm
CCWltrOl
+--as--
Activity
l’-OH - bufuralol
2.0 nmobmg-1. h-’
wt MI Mu Mm Control
FIG.
5. Expression of single mutations of exon 1 (MI) and
exon 2
(M11, MIII) of
the CYP2D6 gene.
The mutations indi-
cated in the top
panel
were introduced into the wild-type (wt) IIDG-
cDNA as described under “Materials and Methods,” transiently ex-
pressed in CO&l cells, and the expressed protein analyzed by West-
ern blotting and bufuralol 1’-hydroxylation activity as in Fig.
4.
described above. Therefore the three amino acid changes in
the 5’ part of the gene together or alone are capable of
destroying the function of this protein. However, chimeric
gene 4 conferred on expression the same or even higher
activity as the wild-type gene. This indicates that the amino
acid change (486 Ser to Thr) caused by the mutation in the
9th exon is not important for expression or activity. Western
blot analysis of the products of clones 2, 3, and 4 revealed
several additional shorter bands, some of which were also
seen in the products of the wild-type gene (lane 2).
Mutations in Exons 1 and 2-The three mutations causing
amino acid changes in exons 1 and 2, which were expressed
in combination in chimeric gene 3, were reproduced separately
by site-specific mutation of the IIDG-cDNA and expressed in
COS-1 cells (Fig. 5). All three mutated cDNAs resulted in
immunoreactive protein, but only the mutation in exon 1(188
C to T, 34 Pro to Ser) abolished the activity of the expressed
protein, the activity being as low as in mock-transfected
control cells.
The expression experiment described in Fig. 4 was per-
formed at least 3 times (chimeras 3 and 4) and up to 6 times
(chimeras 1 and 2). The same results in regard to relative
activities and proteins on Western blots were observed. Two
experiments with the cDNAs in which the mutations of exon
1 and 2 were introduced (Fig. 5) also revealed reproducibility
of the reported findings.
DISCUSSION
In the present report we describe multiple mutations of the
human CYP2D6 gene in two types of mutant alleles isolated
from PMs of debrisoquine. A causal relationship between one
of these mutations, a point mutation in the splice site consen-
sus sequence, and the previously shown absence of cytochrome

Debrisoquine Polymorphism
17213
P450IID6 in livers of PMs (4, 5) is strongly suggested. This
was possible by functional expression in cell culture of chi-
merit genes in which parts of the mutant gene were combined
with parts of the wild-type gene. Our studies provide an
explanation at the DNA level for the previously postulated
mechanisms of aberrant splicing of P450IID6 premRNA (5)
and define two additional mutant alleles of the CYP2D6 gene
associated with the PM phenotype.
The two types of mutant CYP2D6 alleles described here,
designated 29-A and 29-B, were isolated from genomic DNA
libraries of three PM individuals. They appear to occur at
different frequencies. The 29-A allele, which contains a single
frameshift mutation in exon 5 (Fig. 3), was identified as one
of two alleles in only one individual, whereas the 29-B allele
was found in all three individuals. These data suggest that
the 29-B allele may be a common cause of deficient metabo-
lism of debrisoquine, because -75% of PM individuals have
at least one 29-kb fragment of X&I Southern blots (9).
Obviously, studies in a larger number of phenotyped individ-
uals are necessary to evaluate this point.
The two previously identified mutant alleles characterized
by X&I 44- and 11.5-kb fragments occurred with allele fre-
quencies of 0.31 and 0.14, respectively, in PM individuals (9).
The mutations in these alleles which lead to the PM pheno-
type are not known yet, The two new alleles described here
bring the total number of mutant alleles to at least four. We
have accumulated preliminary evidence that these four alleles
account for the great majority of variant CYP2D6 genes
associated with poor metabolism of debrisoquine. Moreover,
this study provides first insights into the molecular mecha-
nism at the DNA level of the debrisoquine polymorphism.
The single mutation in exon 5 of the mutant allele 29-A,
the deletion of one nucleotide, causes a reading frame disrup-
tion and therefore, if present in the homozygous state, readily
would explain the absence of IID protein and function in
poor metabolizers by premature termination of protein syn-
thesis.
The allele 29-B contained four mutations causing amino
acid changes. Only the mutations in the middle part of the
mutant 29-B allele (chimera 1) on expression resulted in a
total absence of the IID protein and consequently its func-
tion, as measured by virtually absent bufuralol l’-hydroxyl-
ation. A total of three mutations were identified in this
segment, namely the 1749 G to C in exon 3, the additional
2185 A to G in intron 4 detected when sequencing the whole
intron, and the 1934 G to A at the 3’ end of intron 3. We
suspect that the G to A change at the last nucleotide of intron
3 found in all 29-B alleles is the dominant cause for absent
protein and function, because the consensus acceptor site
sequence AG is conserved to 100% in numerous genes of
human and other species examined (25, 26). Point mutagen-
esis experiments support the concept that the “AG” consensus
acceptor site sequence is a prerequisite for a normal splicing
mechanism (27). Interestingly, there was no difference in the
size of the RNA in the Northern blots analysis of the COS
cells in which this mutation was expressed (Fig. 4B), which
would point to retention of an intron or truncation of the
mRNA, but it is possible that too small a number of nucleo-
tides was deleted or retained by aberrant splicing to be de-
tected by this technique. In any case, the present observations
support our previous proposal that aberrant splicing may be
a cause of absent IID protein in livers of PMs of debrisoquine
(5). In these studies, Northern blot analysis with RNA from
livers of PMs suggested the presence of additional RNA
bands. The origin of these additional bands was suspected to
come from retained introns 5 and 6, because cDNA sequences
from some of these PM livers contained these intronic se-
quences. The present data therefore may suggest that a defect
in the splice site consensus sequence for the third intron may
lead to retention of other introns, a hypothesis presently
under investigation.
The three amino acid changes in the 5’ part of the 29-B
allele were tested separately by site-specific mutations in
regard to their consequences for protein synthesis, stability,
or function. These mutations, in combination (chimera 3) led
to a total loss of P450IID6 function on expression, although
immunoreactive protein of the right molecular weight was
formed. The experiments summarized in Fig. 5 reveal that
the mutation in exon 1 is predominantly responsible for this
change in function. The amino acid changes in exons 2 (Fig.
5) and 9 (chimera 4, Fig. 4B) had no significant effect on
either IID protein or function. In fact the activity and
amount of protein in four experiments with chimera 4 always
appeared higher when compared with the expression of the
wild-type gene, but the semiquantitative nature of these ex-
periments precludes further interpretation. Thus, these mu-
tations have no apparent significance for the PM trait. To-
gether the above experiments demonstrate that the 29-B allele
has at least two mutations (the G to A splice site mutation in
intron 3 and the C to T point mutation in exon l), either of
which can abolish the function of P450IID6.
Several additional protein bands of faster mobility and
presumably smaller size were observed on expression of the
chimeric genes, but the same bands were also present in COS
cells in which the normal wild-type CYP2D6 was expressed
(Fig. 4). Expression of the full length IID cDNA in COS-1
cells under the same conditions always produced only a single
protein band recognized by the same monoclonal antibody as
the one used in the present study (Fig. 5). These additional
bands also were not observed in livers of PMs so far. Thus,
the expression of the entire gene and of the cDNA produces
different results at least in the COS-1 cell system used. The
additional bands could reflect lability of the protein or alter-
nate splicing mechanisms and have not yet been investigated
further.
In trying to explain the highly polymorphic nature of the
CYP2D locus noted on restriction analysis of genomic DNA
in a larger population (9) and further documented here, one
observation of the present study is of particular interest. With
exception of the mutation in the splice site and in exon 1, the
29-B allele contains the same mutations and the additional
BamHI restriction site as the recently reported CYP2D7 gene,
which is located 5’ of CYP2D6 and suspected to represent a
gene duplication of CYP2D6, having 97% amino acid similar-
ity to CYP2D6 (8). It is not yet entirely clear whether
CYP2D7 is transcribed into a mRNA and produces a protein
as it has only a single reading frame disrupting insertion in
its 1st exon (8). The shared mutations between CYP2D7 and
mutated CYP2D6 genes, as reflected by the 29-B allele, may
be a consequence of homologous recombination or gene con-
version events. Gene conversion between CYP2D7 and a
pseudogene CYP2D8, which is located 5’ of CYP2D7, have
already been observed (8). Gene conversions also have been
described in several other P-450 families. For instance, gene
conversions between the CYP21A2 and the neighboring
CYPBlAl pseudogene are known to contribute to the muta-
tions at the steriod 21-hydroxylase locus (28).
The CYP2D6 gene probably has no important physiological
function or is not essential for present day life. Many muta-
tions therefore may have accumulated even in a still func-
tional gene. Gene conversion events contribute to this accu-
mulation of mutations by spreading them in this gene family.

17214 Debrisoquine
In the course of the accumulation of mutations, splicing errors
presumably will appear sooner or later. Alternative splicing
mechanisms or aberrant splicing has been reported for other
cytochrome P-450 enzymes (28, 29). It is conceivable that
more splicing errors
will be found in drug metabolizing P-450
genes, because they are not under selective pressure. Even
normally functioning genes of this group of enzymes may
have
fragile splicing mechanisms and structures. We believe
that the polymorphic CYP2D6 gene might be a gene predis-
posed to extinction. Both the mechanisms of gene conversion
and of aberrant splicing seem to play an important role in
these events. Many inactivated or P-450 pseudogenes will
probably be found in the human genome.
On the other hand, these polymorphic genes play an im-
portant role as causes of interindividual variation in drug
metabolism and in the occurrence of side effects and thera-
peutic failures. Moreover, they serve as genetic markers for
numerous diseases. The elucidation of these mutations there-
fore has clinical importance and the definition of the muta-
tions of the IID gene will allow the development of simple
tests for the detection of the respective genotype.
Acknowledgments-The editorial assistance of Marianne Liechti
and the technical assistance of Therese Catin are greatly appreciated.
We also thank Donald J. Birkett for helpful discussions and review
of the manuscript.
REFERENCES
1. Meyer, U. A., Skoda, R. C., and Zanger, U. M. (1990) Phurmacol.
Ther. 46,297-308
2. Eichelbaum, M. (1988) ZSZ Atlas Sci. Pharmacol. 243-251
3. Kalow, W., Goedde, H. W., and Agarwal, D. P. (1986) Ethnic
Differences in Reactions to Drugs and Xenobiotics, pp. 157-208,
Alan R. Liss, New York
4. Zanger, U. M., Vilbois, F., Hardwick, J. P., and Meyer, U. A.
(1988) Biochemistry 27, 5447-5454
5. Gonzalez, F. J., Skoda, R. C., Kimura, S., Umeno, M., Zanger, U.
M., Nebert, D. W., Gelboin, H. V., Hardwick, J. P., and Meyer,
U. A. (1988) Nature 331,442-446
6. Nebert, D. W., Nelson, D. R., Adesnik, M., Coon, M. J., Esta-
brook, R. W., Gonzalez, F. J., Guengerich, F. P., Gunsalus, I.
Polymorphism
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
C., Johnson, E. F., Kemper, B., Levin, W., Phillips, I. R., Sato,
R., and Waterman, M. R. (1989) DNA (N. Y.) 8, 1-13
Gonzalez, F. J., Vilbois, F., Hardwick, J. P., McBride, 0. W.,
Nebert, D. W., Gelboin, H. V., and Meyer, U. A. (1988) G’e-
nomics 2,174-179
Kimura, S., Umeno, M., Skoda, R. C., Gelboin, H. V., Meyer, U.
A., and Gonzalez, F. J. (1989) Am. J. Hum. Genet. 45,889-904
Skoda, R. C., Gonzalez, F. J., Demierre, A., and Meyer, U. A.
(1988) Proc. N&l. Acad. Sci. U. S. A. 85, 5240-5243
Meier, P. J., Mueller, H. K., Dick, B., and Meyer, U. A. (1983)
Gastroenterology 85,682-692
Eichelbaum, M., Reetz, K.-P., Schmidt, E. K., and Zekorn, C.
(1986) Xenobiotica 16,465-481
Kaaimoto, M., Winter, J. S. D.. Kanimoto. K.. Simpson, E. R.,
and Waterman, M. k. (1988) Mol.-Endocn’no~. 2, i64-,70
Saneer. F.. Nicklen. S.. and Coulson. A. R. (1977) Proc. Natl.
A&d: SC;. U. S. A: 7k, 5463-5467
Hattori, M., and Sakaki, Y. (1986) Anal. Biochem. 152, 232-238
Andersson, S., Davis, D. L., Dahlback, H., Jornvall, H., and
Russell, D. W. (1989) J. Biol. Ckem. 264,8222-8229
Thomson, D. R., Stenberg, R. M., Goins, W. F., and Stinski, M.
F. (1984) Proc. N&l. Acad. Sci. U. S. A.
81,659-663
Kammann, M., Laufs, J., Schell, J., and Gronenborn, B. (1989)
Nucleic Acids Res.
17, 5404
Gluzman, Y. (1981) Cell 23, 175-182
Sompavrac, L. M.. and Danna, K. J. (1981) Proc. N&l. Acad. Sci.
U.-S.-A. i8, 7575-7578
Zuber. M. X.. Simuson. E. R.. and Waterman. M. R. (1986)
Science 234,1258-1261
Zanger, U. M., Hauri, H. P., Loeper, J., Homberg, J. C., and
Meyer, U. A. (1988) Proc. Natl. Acad. Sci. U. S. A. 85, 8256-
8260
Wilkinson, M. (1988) Nucleic Acids Res.
16,
10933
Maniatis, T., Fritsch, E. F., and Sambrook, J. (1982) Molecular
Cloning: A Laboratory Manual, pp. 202-203, Cold Spring Har-
bor Laboratory, Cold Spring Harbor, NY
Feinberg, A. P., and Vogelstein, B. (1983) Anal. Biochem.
132,
6-13
Mount, S. M. (1982) Nucleic Acids Res.
10,459-472
Ohshima. Y.. and Gotoh. Y. (1987) J. Mol. Biol.
195, 247-259
Aebi, M.,‘Hdrnig, H., Padgett, R. A., Reiser, J., and Weissmann,
C. (1986) Cell 47,555-565
Higashi, Y., Tanae, A., Inoue, H., Hiromasa, T., and Fujii-Kuri-
yama, Y. (1988) Proc. Natl. Acad. Sci. U. S. A. 85, 7486-7490
Kimura, H., Sogawa, K., Sakai, Y., and Fujii-Kuriyama, Y. (1989)
J. BioE. Chem. 264,2338-2342